Autoregressive LSTM-Based Mobile Energy Management System Considering Spatiotemporal Solar Irradiance Prediction along Travel Routes

With growing interest in renewable energy utilization, artificial intelligence techniques have been widely applied to photovoltaic power forecasting. However, most previous studies have focused on fixed PV systems and do not reflect the characteristics of PV panels mounted on moving vehicles. This paper proposes a power generation forecasting method for vehicle-mounted PV systems using an autoregressive Long Short-Term Memory model. Driving route information was collected through a navigation API, while weather data were obtained from the Korea Meteorological Administration API. The proposed method predicts solar power generation according to vehicle routes and operating conditions. Simulations using electric bus operation data from Jeju Island demonstrate the feasibility of improving mobile PV power prediction and vehicle energy management efficiency.